Research & Intelligence
Original proposed methodologies and findings from our threat intelligence team.
AI securityassessmentmethodology
A Proposed Evidence-Led Assessment Method for AI Systems
An original, practical method for assessing AI risk from intended use through architecture, evaluation, and operational evidence.
By Alexa Cybersecurity Editorial Team9/21/2026
Read Paper MCPreference architectureassessment
A Reference Architecture and Review Method for Secure MCP Use
A proposed trust-zone architecture and review sequence for Model Context Protocol clients, servers, and tools.
By Alexa Cybersecurity Editorial Team9/21/2026
Read Paper AI agentschecklistsecure deployment
A Pre-Deployment Checklist for AI Agents with Tool Access
A proposed release checklist that tests an agent's delegated authority, failure containment, and operational evidence.
By Alexa Cybersecurity Editorial Team9/21/2026
Read Paper